OSM report 2022-2023
Marissa Dyck
2024-03-25
Before you begin
Notes
A few notes about this script.
If you are running this with the 2022-2023 data make sure you download the whole (OSM_2022-2023 GitHub repository)[https://github.com/ACMElabUvic/OSM_2022-2023] from the ACMElabUvic GitHub. This will ensure you have all the files, data, and proper folder structure you will need to run this code and associated analyses.
Also make sure you open RStudio through the R project (OSM_2022-2023.Rproj) this will automatically set your working directory to the correct place (wherever you saved the repository) and ensure you don’t have to change the file paths for some of the data.
Lastly, if you are looking to adapt this code for a future year of data, you will want to ensure you have run the ACME_camera_script_9-2-2024.R or .Rmd with your data as there is much data formatting, cleaning, and restructuring that has to be done before this code will work.
If you have question please email the most recent author, currently
Marissa A. Dyck
Postdoctoral research fellow
University of Victoria
School of Environmental Studies
Email: marissadyck17@gmail.com
Install packages
If you don’t already have the following packages installed, use the code below to install them.
install.packages('tidyverse')
install.packages('ggpubr')
install.packages('corrplot')
install.packages('Hmisc')
install.packages('glmmTMB')
install.packages('MuMIn')Load libraries
Then load the packages to your library.
library(tidyverse) # data tidying, visualization, and much more; this will load all tidyverse packages, can see complete list using tidyverse_packages()
library(ggpubr) # make modificaions to plot for publication (arrange plots)
library(PerformanceAnalytics) #Used to generate a correlation plot## Loading required package: xts## Loading required package: zoo##
## Attaching package: 'zoo'## The following objects are masked from 'package:base':
##
## as.Date, as.Date.numeric##
## ################################### WARNING ###################################
## # We noticed you have dplyr installed. The dplyr lag() function breaks how #
## # base R's lag() function is supposed to work, which breaks lag(my_xts). #
## # #
## # Calls to lag(my_xts) that you enter or source() into this session won't #
## # work correctly. #
## # #
## # All package code is unaffected because it is protected by the R namespace #
## # mechanism. #
## # #
## # Set `options(xts.warn_dplyr_breaks_lag = FALSE)` to suppress this warning. #
## # #
## # You can use stats::lag() to make sure you're not using dplyr::lag(), or you #
## # can add conflictRules('dplyr', exclude = 'lag') to your .Rprofile to stop #
## # dplyr from breaking base R's lag() function. #
## ################################### WARNING #####################################
## Attaching package: 'xts'## The following objects are masked from 'package:dplyr':
##
## first, last##
## Attaching package: 'PerformanceAnalytics'## The following object is masked from 'package:graphics':
##
## legendlibrary(Hmisc) # used to generate histograms for all variables in data frame## Loading required package: lattice## Loading required package: survival## Loading required package: Formula##
## Attaching package: 'Hmisc'## The following objects are masked from 'package:dplyr':
##
## src, summarize## The following objects are masked from 'package:base':
##
## format.pval, unitslibrary(glmmTMB) #Constructing GLMMs## Warning in checkMatrixPackageVersion(): Package version inconsistency detected.
## TMB was built with Matrix version 1.4.1
## Current Matrix version is 1.5.3
## Please re-install 'TMB' from source using install.packages('TMB', type = 'source') or ask CRAN for a binary version of 'TMB' matching CRAN's 'Matrix' package## Warning in checkDepPackageVersion(dep_pkg = "TMB"): Package version inconsistency detected.
## glmmTMB was built with TMB version 1.9.6
## Current TMB version is 1.9.1
## Please re-install glmmTMB from source or restore original 'TMB' package (see '?reinstalling' for more information)library(MuMIn) # for model selectionData
Load detection data
Read in saved and cleaned detection data from the ACME_camera_script_9-2-2024.R.
# detection data
# read in saved and cleaned detection data from the ACME_camera_script_9-2-2024.R
detections <- read_csv('data/processed/OSM_2022_ind_det.csv') %>%
# change site, species and event_id to factor
mutate_if(is.character,
as.factor)## Rows: 14102 Columns: 8
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr (4): array, site, species, event_id
## dbl (3): month, year, timediff
## dttm (1): datetime
##
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.Data formatting
In order to get plots that have the same formatting as last years’ report we have to do a bit of data formatting. First we need to make sure we are including the same relevant species (some were ignored for last years’ report or grouped together)
Last years report had the following species
- white-tailed deer
- snowshoe hare
- black bear
- coyote
- red squirrel
- fisher
- unknown
- moose
- lynx
- spruce grouse
- red fox
- striped skunk
- ruffed grouse
- owl
- grey wolf
- domestic dog
- cougar
- raven
- other
- mule deer
And they grouped all humans except for staff as ‘Humans’. Let’s look at the species we have in this year’s data and try to format it the same way
detections %>%
# group by array and species
group_by(array, species) %>%
summarise(n = n()) %>%
# have R print everything
print(n = nrow(.))## `summarise()` has grouped output by 'array'. You can override using the
## `.groups` argument.## # A tibble: 119 × 3
## # Groups: array [4]
## array species n
## <fct> <fct> <int>
## 1 LU01 Beaver 1
## 2 LU01 Black bear 380
## 3 LU01 Cougar 7
## 4 LU01 Coyote 581
## 5 LU01 Domestic dog 6
## 6 LU01 Fisher 111
## 7 LU01 Grey jay 14
## 8 LU01 Grey wolf 21
## 9 LU01 Human 3
## 10 LU01 Lynx 55
## 11 LU01 Moose 99
## 12 LU01 Other 1
## 13 LU01 Other birds 60
## 14 LU01 Otter 2
## 15 LU01 Owl 2
## 16 LU01 Porcupine 5
## 17 LU01 Raven 6
## 18 LU01 Red fox 50
## 19 LU01 Red squirrel 879
## 20 LU01 Ruffed grouse 14
## 21 LU01 Short-tailed weasel 5
## 22 LU01 Snowshoe hare 1443
## 23 LU01 Spruce grouse 12
## 24 LU01 Staff 71
## 25 LU01 Striped skunk 39
## 26 LU01 Unknown 210
## 27 LU01 Unknown canid 48
## 28 LU01 Unknown deer 175
## 29 LU01 Unknown mustelid 13
## 30 LU01 Unknown ungulate 8
## 31 LU01 White-tailed deer 1953
## 32 LU13 ATVer 31
## 33 LU13 Black bear 275
## 34 LU13 Caribou 3
## 35 LU13 Coyote 187
## 36 LU13 Fisher 5
## 37 LU13 Grey jay 2
## 38 LU13 Grey wolf 52
## 39 LU13 Human 2
## 40 LU13 Hunter 1
## 41 LU13 Long-tailed weasel 1
## 42 LU13 Lynx 115
## 43 LU13 Marten 27
## 44 LU13 Moose 128
## 45 LU13 Other birds 12
## 46 LU13 Owl 1
## 47 LU13 Red fox 2
## 48 LU13 Red squirrel 240
## 49 LU13 Ruffed grouse 7
## 50 LU13 Short-tailed weasel 7
## 51 LU13 Snowshoe hare 573
## 52 LU13 Spruce grouse 25
## 53 LU13 Staff 82
## 54 LU13 Striped skunk 1
## 55 LU13 Unknown 86
## 56 LU13 Unknown canid 10
## 57 LU13 Unknown deer 5
## 58 LU13 Unknown mustelid 3
## 59 LU13 White-tailed deer 86
## 60 LU13 Wolverine 8
## 61 LU15 ATVer 1
## 62 LU15 Beaver 2
## 63 LU15 Black bear 220
## 64 LU15 Canada goose 3
## 65 LU15 Caribou 51
## 66 LU15 Coyote 171
## 67 LU15 Fisher 25
## 68 LU15 Grey jay 21
## 69 LU15 Grey wolf 61
## 70 LU15 Long-tailed weasel 15
## 71 LU15 Lynx 122
## 72 LU15 Marten 63
## 73 LU15 Moose 157
## 74 LU15 Other birds 59
## 75 LU15 Otter 5
## 76 LU15 Owl 1
## 77 LU15 Red fox 39
## 78 LU15 Red squirrel 643
## 79 LU15 Ruffed grouse 11
## 80 LU15 Short-tailed weasel 7
## 81 LU15 Snowmobiler 1
## 82 LU15 Snowshoe hare 611
## 83 LU15 Spruce grouse 21
## 84 LU15 Staff 78
## 85 LU15 Unknown 98
## 86 LU15 Unknown canid 7
## 87 LU15 Unknown deer 47
## 88 LU15 Unknown mustelid 16
## 89 LU15 Unknown ungulate 5
## 90 LU15 White-tailed deer 429
## 91 LU21 Black bear 544
## 92 LU21 Canada goose 1
## 93 LU21 Caribou 16
## 94 LU21 Cougar 2
## 95 LU21 Coyote 51
## 96 LU21 Fisher 46
## 97 LU21 Grey jay 13
## 98 LU21 Grey wolf 55
## 99 LU21 Long-tailed weasel 1
## 100 LU21 Lynx 72
## 101 LU21 Marten 50
## 102 LU21 Moose 233
## 103 LU21 Other 1
## 104 LU21 Other birds 44
## 105 LU21 Owl 8
## 106 LU21 Red fox 14
## 107 LU21 Red squirrel 219
## 108 LU21 Ruffed grouse 11
## 109 LU21 Short-tailed weasel 2
## 110 LU21 Snowmobiler 6
## 111 LU21 Snowshoe hare 284
## 112 LU21 Spruce grouse 19
## 113 LU21 Staff 71
## 114 LU21 Unknown 162
## 115 LU21 Unknown canid 5
## 116 LU21 Unknown deer 65
## 117 LU21 Unknown mustelid 23
## 118 LU21 Unknown ungulate 4
## 119 LU21 White-tailed deer 839# now let's create a new data frame (tibble) to work with for the OSM figure summaries specifically
# I personally would lump all the unknown together and all the birds together but for the sake of consistency with last years' figures we will remove some entries, let's create a vector of entries to drop
species_drop <- c('Staff',
'Unknown deer',
'Unknown ungulate',
'Unknown canid',
'Unknown mustelid',
'Other birds')
# now we can create the new data frame with some changes consistent w/ choices made for 2021-2022
detections <- detections %>%
# for summarizing, lets lump all the recreational humans into "Humans"
mutate(species = recode_factor(species,
"Snowmobiler" = "Human",
"ATVer" = "Human",
'Hunter' = 'Human')) %>%
# remove species we don't want to plot
filter(!species %in% species_drop)We will also want to subset the data by landscape unit (LU) and generate a new data frame for each LU to use for plotting
# we will also want to create a data frame for each LU to plot individually
# LU1
dets_LU1 <- detections %>%
filter(array == 'LU01')
# LU13
dets_LU13 <- detections %>%
filter(array == 'LU13')
# LU15
dets_LU15 <- detections %>%
filter(array == 'LU15')
# LU21
dets_LU21 <- detections %>%
filter(array == 'LU21')ANDREW HELP
Can you make the above code into a forloop which assigns each new data frame created from subsetting as dets_LUname?
Detection plots
Detection data
Now we can apply the same data formatting for each LUs’ data frame using purrr.
We want to count the number of independent detections per species per LU to use in the detection plots
# apply the same formatting to each LU data frame using purrr map
detection_data <- list(dets_LU1,
dets_LU13,
dets_LU15,
dets_LU21) %>%
purrr::map(
~.x %>%
# group by species
group_by(species) %>%
# calculate a column with unique accounts of each species
mutate(count = n_distinct(event_id)) %>%
# keep just the columns we need
select(species, count) %>%
# keep only unique (distinct) rows so we should be left with one row per species, this helps with plotting later if you don't do it ggplot will try to count and plot each row it's annoying
distinct()) %>%
# set names of list objects
purrr::set_names('Detections LU01',
'Detections LU13',
'Detections LU15',
'Detections LU21')Detection plots
Now to graph independent detections for each LU using purrr, this avoids a TON of code repetition needed to plot each one individually
We use purrr::imap() instead of
purrr::map() because imap maintains the variable names in
our list (e.g. Detections LU01, Detections LU13, etc.) which we can then
use to title each plot.
Within purrr::imap() we just paste the code we would use
for a single ggplot since all the graphical elements (except the title
which we change with the file name [.y]) are the same
# create object detection plots which uses the detection_data list (w/ all 4 LUs)
detection_plots <- detection_data %>%
# use imap instead of map as it allows us to use .y to paste the list element names as the plot titles later
purrr::imap(
~.x %>%
# now just copy and paste the ggplot code for the detection graphs
ggplot(.,
aes(x = reorder(species, count), y = count)) +
# plot as bar graph using geom_col so we don't have to provide a y aesthetic
geom_col() +
# switch the x and y axis
coord_flip() +
# add the number of detections at the end of each bar
geom_text(aes(label = count),
color = "black",
size = 3,
hjust = -0.3,
vjust = 0.2) +
# label x and y axis with informative titles
labs(x = 'Species',
y = 'Number of Independent (30 min) Detections') +
# add title to plot with LU name the .y will take the name of whatever you named each list element in the detection_data list, so make sure this name is what you want on the ggtitle
ggtitle(.y) +
# set the theme
theme_classic() +
theme(plot.title = element_text(hjust = 0.5)))
# view plots, this will print each in it's own window so you have to scroll back in the plot viewer pane to look at each one
detection_plots## $`Detections LU01`
##
## $`Detections LU13`
##
## $`Detections LU15`
##
## $`Detections LU21`
Save detection plots
Now we want to save these plots in case we need each individual one (we will combine the detection and naive occ plots into a single figure for each LU later and use those for the OSM report, but we may want these standalone plots later so let’s save them while they are here).
We can save all the plots from the purrr iteration above using
purrr::imap. imap is used instead of map because it allows
us to retain the list object names (plot names) to paste as the file
name with the .y command.
IMPORTANT if you are using this code for a future github repo, DO NOT use .tiff as the file extension. This will cause issues when trying to push any changes to the github repo as the files are too large to meet githubs requirements
# save plots only use if needed
purrr::imap(
detection_plots,
~ggsave(.x,
file = paste0("figures/",
.y,
'.jpg'), # avoid using .tiff extension in the github repo, those files are too large to push to origin
dpi = 600,
width = 11,
height = 9,
units = 'in'))## $`Detections LU01`
## [1] "figures/Detections LU01.jpg"
##
## $`Detections LU13`
## [1] "figures/Detections LU13.jpg"
##
## $`Detections LU15`
## [1] "figures/Detections LU15.jpg"
##
## $`Detections LU21`
## [1] "figures/Detections LU21.jpg"Naive occupancy
Data
We also need to alter the detection data a bit to use for naive occupancy plots.
We will use the individual LU detection data like we did before and
use purrr::map() to apply the dame data formatting to all 4
data frames.
Here we want to calculate the total number of sites in each LU, the number of sites each species was detected at in each LU and then use both those numbers to calculate naive occupancy for each species in each LU
# First we need to alter the data frame a bit for these plots, let's create a data frame for each LU (I couldn't figure out how to do this without assigning individual data frames for each UGH)
# apply the same formatting to each data frame using purrr
occupancy_data <- list(dets_LU1,
dets_LU13,
dets_LU15,
dets_LU21) %>%
purrr::map(
~.x %>%
# calculate the total number of sites for each LU
mutate(total_sites = n_distinct(site)) %>%
# group by species to calculate the number of sites each spp occurred at
group_by(species) %>%
# add columns to count the number of sites each spp occurred at and then the naive occupancy
reframe(count = n_distinct(site),
naive_occ = count/total_sites,
ind_det = n_distinct(event_id)) %>%
# keep just the columns we need
select(species, naive_occ, ind_det) %>%
# keep only unique (distinct) rows so we should be left with one row per species, this helps with plotting
distinct()) %>%
purrr::set_names('Naive Occupancy LU01',
'Naive Occupancy LU13',
'Naive Occupancy LU15',
'Naive Occupancy LU21')Occupancy plots
Now we can graph naive occupancy for each LU using purrr, and as with the detection plots this saves a massive amount of coding using purrr to run an iteration on the data files and produce four plots at once instead of copying and pasting code for each individually
# create object occupancy_plots which uses the occupancy_data list (w/ all 4 LUs)
occupancy_plots <- occupancy_data %>%
# use imap instead of map as it allows us to use .y to paste the list element names as the plot titles later
purrr::imap(
~.x %>%
# now just copy and paste the ggplot code for the occupancy graphs
ggplot(.,
aes(x = fct_reorder(species,
ind_det), # this reorders the species so they match the order of the detection plot which makes it better for viewing when the plots are arranged together in 1 figure for each LU
y = naive_occ)) +
# plot as bars using geom_col() which uses stat = 'identity', instead of geom_bar() which will count the rows in each group and plot that instead of naive occ
geom_col() +
# flip x and y axis
coord_flip() +
# add text to end of bars that provides naive occ value
geom_text(aes(label = round(naive_occ, 2)),
size = 3,
hjust = -0.3,
vjust = 0.2) +
# relabel x and y axis and title
labs(x = 'Species',
y = 'Proportion of Sites With At Least One Detection') +
# set plot title using .y (name of list object)
ggtitle(.y) +
# set. theme elements
theme_classic()+
theme(plot.title = element_text(hjust = 0.5)))
# view plots
occupancy_plots## $`Naive Occupancy LU01`
##
## $`Naive Occupancy LU13`
##
## $`Naive Occupancy LU15`
##
## $`Naive Occupancy LU21`
Save occupancy plots
As with the detection plots, we might want these individual plots
later for something so we can use purrr::imap() to save
them to the figures folder
Again avoid using the .tiff extension in github
# save plots
purrr::imap(
occupancy_plots,
~ggsave(.x,
file = paste0("figures/",
.y,
'.jpg'), # avoid using .tiff extension in the github repo, those files are too large to push to origin
dpi = 600,
width = 11,
height = 9,
units = 'in'))## $`Naive Occupancy LU01`
## [1] "figures/Naive Occupancy LU01.jpg"
##
## $`Naive Occupancy LU13`
## [1] "figures/Naive Occupancy LU13.jpg"
##
## $`Naive Occupancy LU15`
## [1] "figures/Naive Occupancy LU15.jpg"
##
## $`Naive Occupancy LU21`
## [1] "figures/Naive Occupancy LU21.jpg"Final combined plots for report
The previous year’s report had a figure for each LU with the
detections plot on the top and the occupancy plot on the bottom so we
will recreate these for this year using ggarrange().
Unfortunately I could not figure out how to do this in purrr to reduce coding but luckily it isn’t too much repitition
# not sure I know how to do the following section in purrr just yet, but we've saved a ton of coding so far and it doesn't take much to arrange each of these individually
# LU1
# arrange the plots so each LU has a figure with detections on top and naive occ on bottom
LU1_det_occ_plots <- ggarrange(detection_plots$`Detections LU01`, occupancy_plots$`Naive Occupancy LU01`,
labels = c("A", "B"),
nrow = 2)
# view plot
LU1_det_occ_plots
# LU13
# arrange the plots so each LU has a figure with detections on top and naive occ on bottom
LU13_det_occ_plots <- ggarrange(detection_plots$`Detections LU13`, occupancy_plots$`Naive Occupancy LU13`,
labels = c("A", "B"),
nrow = 2)
# view plot
LU13_det_occ_plots
# LU15
# arrange the plots so each LU has a figure with detections on top and naive occ on bottom
LU15_det_occ_plots <- ggarrange(detection_plots$`Detections LU15`, occupancy_plots$`Naive Occupancy LU15`,
labels = c("A", "B"),
nrow = 2)
# view plot
LU15_det_occ_plots
# LU21
# arrange the plots so each LU has a figure with detections on top and naive occ on bottom
LU21_det_occ_plots <- ggarrange(detection_plots$`Detections LU21`, occupancy_plots$`Naive Occupancy LU21`,
labels = c("A", "B"),
nrow = 2)
# view plot
LU21_det_occ_plots
We can however, save all the figures again using purrr
# save all figures at once using purrr
final_det_occ_plots <- list(LU1_det_occ_plots,
LU13_det_occ_plots,
LU15_det_occ_plots,
LU21_det_occ_plots) %>%
purrr::set_names('LU01_det_occ_plots',
'LU13_det_occ_plots',
'LU15_det_occ_plots',
'LU21_det_occ_plots') %>%
purrr::imap(
~ggsave(.x,
file = paste0("figures/",
.y,
'.jpg'), # avoid using .tiff extension in the github repo, those files are too large to push to origin
dpi = 600,
width = 12,
height = 15,
units = 'in'))Analysis
Read in data
We need the proportional binomial data and the covariate data (from the ACME_camera_script_9-2-2024.R or .Rmd), let’s read those in now and check the structure of each
# response metric (proportional detections from the from the ACME_camera_script_9-2-2024.R or .Rmd)
prop_detections <- read_csv('data/processed/OSM_2022_proportional_detections.csv')## Rows: 152 Columns: 23
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr (1): site
## dbl (22): black_bear, coyote, fisher, moose, white-tailed_deer, cougar, grey...
##
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.# check variable structure
str(prop_detections)## spc_tbl_ [152 × 23] (S3: spec_tbl_df/tbl_df/tbl/data.frame)
## $ site : chr [1:152] "LU01_06" "LU01_10" "LU01_11" "LU01_13" ...
## $ black_bear : num [1:152] 7 3 4 7 8 9 4 5 7 7 ...
## $ coyote : num [1:152] 4 4 8 10 11 9 11 0 9 4 ...
## $ fisher : num [1:152] 5 3 3 3 2 1 1 1 0 3 ...
## $ moose : num [1:152] 3 2 5 9 1 0 2 4 1 0 ...
## $ white-tailed_deer : num [1:152] 12 5 12 12 13 14 15 9 12 10 ...
## $ cougar : num [1:152] 0 0 1 0 1 0 0 0 0 0 ...
## $ grey_wolf : num [1:152] 0 0 2 0 0 0 1 0 0 0 ...
## $ lynx : num [1:152] 0 0 1 0 1 1 0 0 0 2 ...
## $ red_fox : num [1:152] 0 0 2 0 0 0 0 0 4 0 ...
## $ wolverine : num [1:152] 0 0 0 0 0 0 0 0 0 0 ...
## $ caribou : num [1:152] 0 0 0 0 0 0 0 0 0 0 ...
## $ absent_black_bear : num [1:152] 5 3 8 5 4 3 8 7 5 5 ...
## $ absent_coyote : num [1:152] 10 1 6 5 3 5 4 15 6 11 ...
## $ absent_fisher : num [1:152] 9 2 11 12 12 13 14 14 15 12 ...
## $ absent_moose : num [1:152] 11 3 9 6 13 14 13 11 14 15 ...
## $ absent_white-tailed_deer: num [1:152] 2 0 2 3 1 0 0 6 3 5 ...
## $ absent_cougar : num [1:152] 14 5 13 15 13 14 15 15 15 15 ...
## $ absent_grey_wolf : num [1:152] 14 5 12 15 14 14 14 15 15 15 ...
## $ absent_lynx : num [1:152] 14 5 13 15 13 13 15 15 15 13 ...
## $ absent_red_fox : num [1:152] 14 5 12 15 14 14 15 15 11 15 ...
## $ absent_wolverine : num [1:152] 14 5 14 15 14 14 15 15 15 15 ...
## $ absent_caribou : num [1:152] 14 5 14 15 14 14 15 15 15 15 ...
## - attr(*, "spec")=
## .. cols(
## .. site = col_character(),
## .. black_bear = col_double(),
## .. coyote = col_double(),
## .. fisher = col_double(),
## .. moose = col_double(),
## .. `white-tailed_deer` = col_double(),
## .. cougar = col_double(),
## .. grey_wolf = col_double(),
## .. lynx = col_double(),
## .. red_fox = col_double(),
## .. wolverine = col_double(),
## .. caribou = col_double(),
## .. absent_black_bear = col_double(),
## .. absent_coyote = col_double(),
## .. absent_fisher = col_double(),
## .. absent_moose = col_double(),
## .. `absent_white-tailed_deer` = col_double(),
## .. absent_cougar = col_double(),
## .. absent_grey_wolf = col_double(),
## .. absent_lynx = col_double(),
## .. absent_red_fox = col_double(),
## .. absent_wolverine = col_double(),
## .. absent_caribou = col_double()
## .. )
## - attr(*, "problems")=<externalptr># model covariates (merged HFI and VEG data from the ACME_camera_script_9-2-2024.R or .Rmd)
covariates <- read_csv('data/processed/OSM_2022_covariates.csv',
# set the column types to read in correctly
col_types = cols(array = col_factor(),
camera = col_factor(),
site = col_factor(),
buff_dist = col_factor(),
.default = col_number()))
# check variable structure
str(covariates)## spc_tbl_ [3,100 × 76] (S3: spec_tbl_df/tbl_df/tbl/data.frame)
## $ array : Factor w/ 4 levels "LU13","LU15",..: 1 1 1 1 1 1 1 1 1 1 ...
## $ camera : Factor w/ 96 levels "18","15","03",..: 1 2 3 4 5 6 7 8 9 10 ...
## $ site : Factor w/ 155 levels "LU13_18","LU13_15",..: 1 2 3 4 5 6 7 8 9 10 ...
## $ buff_dist : Factor w/ 20 levels "250","500","750",..: 1 1 1 1 1 1 1 1 1 1 ...
## $ harvest_area : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ crop : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ well_aband : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ well_oil : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ trail : num [1:3100] 0 0 NA 0.5 0 ...
## $ harvest_area_white_zone : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ conventional_seismic : num [1:3100] 0.5 0.5 NA 0.5 1 ...
## $ pipeline : num [1:3100] 0 0.5 NA 0 0 0 0.5 0 0 0 ...
## $ tame_pasture : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ rough_pasture : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ rural_residence : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ transmission_line : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ well_gas : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ misc_oil_gas_facility : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ clearing_unknown : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ vegetated_edge_roads : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ road_unimproved : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ road_gravel_1l : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ road_gravel_2l : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ truck_trail : num [1:3100] 0.5 0 NA 0 0 0 0 0 0 0 ...
## $ borrowpits : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ sump : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ borrowpit_wet : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ cultivation_abandoned : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ urban_residence : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ country_residence : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ recreation : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ well_other : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ well_bitumen : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ well_cased : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ road_paved_undiv_2l : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ road_unclassified : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ runway : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ clearing_wellpad_unconfirmed: num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ facility_unknown : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ borrowpit_dry : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ grvl_sand_pit : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ dugout : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ lagoon : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ open_pit_mine : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ low_impact_seismic : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ surrounding_veg : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ transfer_station : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ facility_other : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ vegetated_edge_railways : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ fruit_vegetables : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ residence_clearing : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ cfo : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ landfill : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ well_cleared_not_confirmed : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ oil_gas_plant : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ urban_industrial : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ road_paved_1l : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ road_paved_undiv_1l : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ road_winter : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ well_cleared_not_drilled : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ well_unknown : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ airp_runway : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ reservoir : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ campground : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ canal : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ camp_industrial : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ rlwy_sgl_track : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ lc_class20 : num [1:3100] 0 0 0.4 0.2 0.5 0 0 0 0 0 ...
## $ lc_class33 : num [1:3100] 0 0 0 0 0 0 0 0 0 0 ...
## $ lc_class34 : num [1:3100] 0 0 0 0 0 0 0 0 0 0 ...
## $ lc_class50 : num [1:3100] 0 0 0.2 0.2 0 0 0 0 0 0 ...
## $ lc_class110 : num [1:3100] 0 0.167 0 0 0 ...
## $ lc_class120 : num [1:3100] 0 0 0 0 0 0 0 0 0 0 ...
## $ lc_class210 : num [1:3100] 0.5 0.333 0.2 0.4 0.5 ...
## $ lc_class220 : num [1:3100] 0.5 0.333 0.2 0.2 0 ...
## $ lc_class230 : num [1:3100] 0 0.167 0 0 0 ...
## - attr(*, "spec")=
## .. cols(
## .. .default = col_number(),
## .. array = col_factor(levels = NULL, ordered = FALSE, include_na = FALSE),
## .. camera = col_factor(levels = NULL, ordered = FALSE, include_na = FALSE),
## .. site = col_factor(levels = NULL, ordered = FALSE, include_na = FALSE),
## .. buff_dist = col_factor(levels = NULL, ordered = FALSE, include_na = FALSE),
## .. harvest_area = col_number(),
## .. crop = col_number(),
## .. well_aband = col_number(),
## .. well_oil = col_number(),
## .. trail = col_number(),
## .. harvest_area_white_zone = col_number(),
## .. conventional_seismic = col_number(),
## .. pipeline = col_number(),
## .. tame_pasture = col_number(),
## .. rough_pasture = col_number(),
## .. rural_residence = col_number(),
## .. transmission_line = col_number(),
## .. well_gas = col_number(),
## .. misc_oil_gas_facility = col_number(),
## .. clearing_unknown = col_number(),
## .. vegetated_edge_roads = col_number(),
## .. road_unimproved = col_number(),
## .. road_gravel_1l = col_number(),
## .. road_gravel_2l = col_number(),
## .. truck_trail = col_number(),
## .. borrowpits = col_number(),
## .. sump = col_number(),
## .. borrowpit_wet = col_number(),
## .. cultivation_abandoned = col_number(),
## .. urban_residence = col_number(),
## .. country_residence = col_number(),
## .. recreation = col_number(),
## .. well_other = col_number(),
## .. well_bitumen = col_number(),
## .. well_cased = col_number(),
## .. road_paved_undiv_2l = col_number(),
## .. road_unclassified = col_number(),
## .. runway = col_number(),
## .. clearing_wellpad_unconfirmed = col_number(),
## .. facility_unknown = col_number(),
## .. borrowpit_dry = col_number(),
## .. grvl_sand_pit = col_number(),
## .. dugout = col_number(),
## .. lagoon = col_number(),
## .. open_pit_mine = col_number(),
## .. low_impact_seismic = col_number(),
## .. surrounding_veg = col_number(),
## .. transfer_station = col_number(),
## .. facility_other = col_number(),
## .. vegetated_edge_railways = col_number(),
## .. fruit_vegetables = col_number(),
## .. residence_clearing = col_number(),
## .. cfo = col_number(),
## .. landfill = col_number(),
## .. well_cleared_not_confirmed = col_number(),
## .. oil_gas_plant = col_number(),
## .. urban_industrial = col_number(),
## .. road_paved_1l = col_number(),
## .. road_paved_undiv_1l = col_number(),
## .. road_winter = col_number(),
## .. well_cleared_not_drilled = col_number(),
## .. well_unknown = col_number(),
## .. airp_runway = col_number(),
## .. reservoir = col_number(),
## .. campground = col_number(),
## .. canal = col_number(),
## .. camp_industrial = col_number(),
## .. rlwy_sgl_track = col_number(),
## .. lc_class20 = col_number(),
## .. lc_class33 = col_number(),
## .. lc_class34 = col_number(),
## .. lc_class50 = col_number(),
## .. lc_class110 = col_number(),
## .. lc_class120 = col_number(),
## .. lc_class210 = col_number(),
## .. lc_class220 = col_number(),
## .. lc_class230 = col_number()
## .. )
## - attr(*, "problems")=<externalptr>Format covariates
There are too many covariates to include in the models individually and many of them describe similar HFI features. We can use the info from the README file in this repository which includes detailed descriptions from the ABMI human footprints wall to wall data download website for Year 2021 OR in the relevant_literature folder of this repository (HFI_2021_v1_0_Metadata_Final.pdf).
the current version of this code for the purposes of the 2022-2023 report used a merged dataset from 2021-2022 and 2022-2023 data, howver each year of data the variables were extracted slightly differenty from GIS so final version of this code will include a different formatting process which will likely occur in the ACME_camera_script_9-2-2024.R or .Rmd
covariates_grouped <- covariates %>%
mutate(borrowpits = rowSums(across(contains('borrowpit'))),
industrial_sites = camp_industrial + oil_gas_plant + open_pit_mine +
rowSums(across(contains('facility'))),
seismic_lines = rowSums(across(contains('seismic'))),
wellsites = rowSums(across(contains('well'))),
roads = rowSums(across(contains('road'))),
havest_areas = rowSums(across(contains('harvest'))),
trails = rowSums(across(contains('trail'))),
residences = rowSums(across(contains('residence'))),
pasture = rowSums(across(contains('pasture'))),
other_transportation_features = runway + airp_runway + rlwy_sgl_track + vegetated_edge_railways,
crops = crop + fruit_vegetables + cultivation_abandoned,
water = lagoon + reservoir + dugout + canal,
.keep = 'unused') %>%
# remove features we don't need
select(!c(recreation,
clearing_unknown,
cfo,
grvl_sand_pit,
transfer_station,
campground,
surrounding_veg,
urban_industrial,
landfill,
sump,
water,
crops,
other_transportation_features,
pasture,
residences
)) %>%
# reorder variables
relocate(c(pipeline,
transmission_line,
borrowpits),
.after = lc_class230)
# see what's left
names(covariates_grouped)## [1] "array" "camera" "site"
## [4] "buff_dist" "lc_class20" "lc_class33"
## [7] "lc_class34" "lc_class50" "lc_class110"
## [10] "lc_class120" "lc_class210" "lc_class220"
## [13] "lc_class230" "pipeline" "transmission_line"
## [16] "borrowpits" "industrial_sites" "seismic_lines"
## [19] "wellsites" "roads" "havest_areas"
## [22] "trails"# check the structure of new data
str(covariates_grouped)## tibble [3,100 × 22] (S3: tbl_df/tbl/data.frame)
## $ array : Factor w/ 4 levels "LU13","LU15",..: 1 1 1 1 1 1 1 1 1 1 ...
## $ camera : Factor w/ 96 levels "18","15","03",..: 1 2 3 4 5 6 7 8 9 10 ...
## $ site : Factor w/ 155 levels "LU13_18","LU13_15",..: 1 2 3 4 5 6 7 8 9 10 ...
## $ buff_dist : Factor w/ 20 levels "250","500","750",..: 1 1 1 1 1 1 1 1 1 1 ...
## $ lc_class20 : num [1:3100] 0 0 0.4 0.2 0.5 0 0 0 0 0 ...
## $ lc_class33 : num [1:3100] 0 0 0 0 0 0 0 0 0 0 ...
## $ lc_class34 : num [1:3100] 0 0 0 0 0 0 0 0 0 0 ...
## $ lc_class50 : num [1:3100] 0 0 0.2 0.2 0 0 0 0 0 0 ...
## $ lc_class110 : num [1:3100] 0 0.167 0 0 0 ...
## $ lc_class120 : num [1:3100] 0 0 0 0 0 0 0 0 0 0 ...
## $ lc_class210 : num [1:3100] 0.5 0.333 0.2 0.4 0.5 ...
## $ lc_class220 : num [1:3100] 0.5 0.333 0.2 0.2 0 ...
## $ lc_class230 : num [1:3100] 0 0.167 0 0 0 ...
## $ pipeline : num [1:3100] 0 0.5 NA 0 0 0 0.5 0 0 0 ...
## $ transmission_line: num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ borrowpits : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ industrial_sites : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ seismic_lines : num [1:3100] 0.5 0.5 NA 0.5 1 ...
## $ wellsites : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ roads : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ havest_areas : num [1:3100] 0 0 NA 0 0 0 0 0 0 0 ...
## $ trails : num [1:3100] 0.5 0 NA 0.5 0 ...# check summary of new data
summary(covariates_grouped)## array camera site buff_dist lc_class20
## LU13:820 27 : 80 LU13_18: 20 250 : 155 Min. :0.00000
## LU15:780 32 : 80 LU13_15: 20 500 : 155 1st Qu.:0.00000
## LU21:720 41 : 80 LU13_03: 20 750 : 155 Median :0.05460
## LU01:780 36 : 80 LU13_34: 20 1000 : 155 Mean :0.07304
## 16 : 60 LU13_57: 20 1250 : 155 3rd Qu.:0.11321
## 21 : 60 LU13_16: 20 1500 : 155 Max. :0.59091
## (Other):2660 (Other):2980 (Other):2170
## lc_class33 lc_class34 lc_class50 lc_class110
## Min. :0.0000000 Min. :0.00000 Min. :0.0000 Min. :0.00000
## 1st Qu.:0.0000000 1st Qu.:0.00000 1st Qu.:0.1000 1st Qu.:0.05882
## Median :0.0000000 Median :0.00000 Median :0.1686 Median :0.12500
## Mean :0.0005196 Mean :0.01445 Mean :0.1829 Mean :0.12674
## 3rd Qu.:0.0000000 3rd Qu.:0.01316 3rd Qu.:0.2500 3rd Qu.:0.17550
## Max. :0.0909091 Max. :0.33333 Max. :1.0000 Max. :0.55556
##
## lc_class120 lc_class210 lc_class220 lc_class230
## Min. :0.00000 Min. :0.0000 Min. :0.0000 Min. :0.00000
## 1st Qu.:0.00000 1st Qu.:0.1064 1st Qu.:0.1726 1st Qu.:0.08217
## Median :0.00000 Median :0.1591 Median :0.2222 Median :0.15152
## Mean :0.03835 Mean :0.1813 Mean :0.2268 Mean :0.15583
## 3rd Qu.:0.00000 3rd Qu.:0.2179 3rd Qu.:0.2735 3rd Qu.:0.22093
## Max. :1.00000 Max. :1.0000 Max. :1.0000 Max. :0.66667
##
## pipeline transmission_line borrowpits industrial_sites
## Min. :0.00000 Min. :0.000000 Min. :0.00000 Min. :0.000000
## 1st Qu.:0.00000 1st Qu.:0.000000 1st Qu.:0.00000 1st Qu.:0.000000
## Median :0.03637 Median :0.000000 Median :0.00000 Median :0.000000
## Mean :0.06940 Mean :0.004712 Mean :0.01108 Mean :0.001448
## 3rd Qu.:0.10638 3rd Qu.:0.000000 3rd Qu.:0.01807 3rd Qu.:0.000000
## Max. :1.00000 Max. :0.500000 Max. :0.16667 Max. :0.111111
## NA's :8 NA's :8 NA's :8 NA's :8
## seismic_lines wellsites roads havest_areas
## Min. :0.0000 Min. :0.00000 Min. :0.00000 Min. :0.00000
## 1st Qu.:0.2774 1st Qu.:0.01541 1st Qu.:0.00000 1st Qu.:0.00000
## Median :0.3868 Median :0.04408 Median :0.05939 Median :0.00000
## Mean :0.4173 Mean :0.05748 Mean :0.15189 Mean :0.04801
## 3rd Qu.:0.5000 3rd Qu.:0.08125 3rd Qu.:0.27978 3rd Qu.:0.04506
## Max. :1.0000 Max. :0.50000 Max. :0.83333 Max. :0.83333
## NA's :8 NA's :8 NA's :8 NA's :8
## trails
## Min. :0.0000
## 1st Qu.:0.0617
## Median :0.1522
## Mean :0.1874
## 3rd Qu.:0.2712
## Max. :1.0000
## NA's :8# there are some NAs in the data which will cause problems with modeling/visualization of data ignore for now but will explore these sites specifically after report
covariates_grouped <- covariates_grouped %>%
# remove rows with NAs
na.omit()Subset data & correlation plots
Marissa try to get the purrr code for this to work later
Now we need to subset the data for each buffer width, and then in the same loop let’s make correlation plots for these variables within each buffer
# Couldn't get this to work in purrr yet so using a loop to subset the data, create the plots, and save them all in one section... NEAT
buffer_frames<-list()
for (i in unique(covariates_grouped$buff_dist)){
print(i)
#Subset data based on radius
df<-covariates_grouped%>%
filter(buff_dist == i)
#rename dataframe on the fly
assign(paste("df", i, sep ="_"), df)
#list of dataframes
buffer_frames<-c(buffer_frames, list(df))
#Subset data based on radius
df<-covariates_grouped%>%
filter(buff_dist == i)%>%
select(where(is.numeric))
#compute a correlation matrix (watch for errors)
matrix<-cor(df)
#print and save the correlation plot on the go
#renaming for each buffer as we do
png(file.path("figures/", paste("correlation_", i, ".png")))
corrplot::corrplot(matrix,
type = 'upper',
tl.col = 'black',
title = paste0('Variable correlation plot at ', i))
dev.off()
}## [1] "250"## Warning in cor(df): the standard deviation is zero## [1] "500"## Warning in cor(df): the standard deviation is zero## [1] "750"## [1] "1000"## [1] "1250"## [1] "1500"## [1] "1750"## [1] "2000"## [1] "2250"## [1] "2500"## [1] "2750"## [1] "3000"## [1] "3250"## [1] "3500"## [1] "3750"## [1] "4000"## [1] "4250"## [1] "4500"## [1] "4750"## [1] "5000"# name list objects so we can extract names for plotting
buffer_frames <- buffer_frames %>%
# absurdly long way to do this but for sake of time fuck it
purrr::set_names('250 meter buffer',
'500 meter buffer',
'750 meter buffer',
'1000 meter buffer',
'1250 meter buffer',
'1500 meter buffer',
'1750 meter buffer',
'2000 meter buffer',
'2250 meter buffer',
'2500 meter buffer',
'2750 meter buffer',
'3000 meter buffer',
'3250 meter buffer',
'3500 meter buffer',
'3750 meter buffer',
'4000 meter buffer',
'4250 meter buffer',
'4500 meter buffer',
'4750 meter buffer',
'5000 meter buffer')Exploratory plots
add more to this section in later when we have more time to explore the covariates and choose which should be inlcuded etc.
hfi_histograms <- buffer_frames %>%
purrr::imap(
~.x %>%
# filter to just the HFI variables
select(where(is.numeric) &
! starts_with('lc_class')) %>%
# pipe into hist.data.frame function to make histograms for each variable
hist.data.frame(mtitl = paste0('Histograms of HFI variables at ', .y)))
Now let’s do the same thing with the landcover variables
lc_histograms <- buffer_frames %>%
purrr::imap(
~.x %>%
# filter to just the landcover variables
select(where(is.numeric) &
starts_with('lc_class')) %>%
# pipe into hist.data.frame function to make histograms for each variable
hist.data.frame(mtitl = paste0('Histograms of landcover variables at ', .y)))
Add response metric
Now that we have the covariate data formatted we need to add the response metric (monthly proportional presence/absence) to the data frames
final_df <- buffer_frames %>%
purrr::map(
~.x %>%
left_join(prop_detections,
by = 'site'))Black bear
ANDREW/MARISSA FIX later
there is probably a way to shorten the following code to select particular species, I saw Andrew’s for loop in the draft script he wrote but couldn’t quite figure it out so I did this instead, maybe we can merge approaches?
Global model
black_bear_mods <- final_df %>%
purrr::map(
~.x %>%
glmmTMB::glmmTMB(cbind(black_bear, absent_black_bear) ~
seismic_lines +
pipeline +
borrowpits +
wellsites +
roads +
trails +
lc_class20 +
lc_class34 +
lc_class50 +
lc_class110 +
lc_class210 +
lc_class220 +
lc_class230 +
(1|array),
data = .,
family = 'binomial'))Model selection
model.sel(black_bear_mods)## Warning in model.sel.default(black_bear_mods): models are not all fitted to the
## same data## Model selection table
## cnd((Int)) dsp((Int)) cnd(brr) cnd(lc_c11) cnd(lc_c20)
## 250 meter buffer -0.8237 + 1.0280 0.40780 0.5002000
## 500 meter buffer -1.8100 + -8.8800 1.48300 2.1650000
## 1000 meter buffer -1.2440 + 5.0040 -0.79470 0.9692000
## 750 meter buffer -0.9882 + 3.2100 0.08085 1.0990000
## 1500 meter buffer -1.2430 + -1.6840 -1.97700 -0.1470000
## 4000 meter buffer -2.5810 + -10.0700 1.06300 2.4720000
## 4250 meter buffer -3.1640 + -12.3600 0.46190 3.0210000
## 1250 meter buffer -1.5400 + -3.2830 -0.87330 0.6002000
## 4500 meter buffer -3.1770 + -11.5900 -0.12570 2.3210000
## 5000 meter buffer -2.9500 + -9.5120 2.01100 3.0470000
## 4750 meter buffer -2.9020 + -10.2100 1.21200 2.9250000
## 3750 meter buffer -2.0630 + -7.5570 -0.28530 1.8680000
## 1750 meter buffer -1.2670 + -2.4830 -1.42300 0.1190000
## 3500 meter buffer -2.1010 + -7.2870 -0.70810 1.3820000
## 3250 meter buffer -0.3758 + -3.9270 -2.37000 0.0878400
## 3000 meter buffer -0.2031 + -3.0550 -2.93400 -0.1205000
## 2750 meter buffer 0.2775 + 0.8066 -2.37500 -0.2794000
## 2250 meter buffer -0.6461 + 2.6130 -1.06400 0.5436000
## 2000 meter buffer -0.7442 + 0.5012 -1.60200 -0.0001237
## 2500 meter buffer -0.1102 + 0.9909 -2.07700 -0.2207000
## cnd(lc_c21) cnd(lc_c22) cnd(lc_c23) cnd(lc_c34) cnd(lc_c50)
## 250 meter buffer -0.02139 0.82310 0.334200 -0.01718 0.05003
## 500 meter buffer 1.02300 2.56100 1.662000 2.70000 1.47400
## 1000 meter buffer -0.99610 0.63190 0.488000 -0.19810 0.25530
## 750 meter buffer -0.68520 1.29900 0.121600 -1.50100 0.01754
## 1500 meter buffer -1.13400 1.20900 -0.657200 -2.59900 -0.55680
## 4000 meter buffer -1.28200 1.08200 3.415000 31.62000 1.35600
## 4250 meter buffer -1.30100 2.05700 3.583000 26.01000 1.64500
## 1250 meter buffer -0.56750 1.28500 0.170100 0.45260 0.06635
## 4500 meter buffer -2.24200 1.93700 3.601000 -1.36700 1.68800
## 5000 meter buffer -1.26500 0.35690 5.558000 24.91000 2.14500
## 4750 meter buffer -1.73800 0.94160 4.978000 22.74000 1.84500
## 3750 meter buffer -1.88700 0.39800 2.642000 11.30000 1.01100
## 1750 meter buffer -1.02700 1.33500 0.063110 4.37900 0.10100
## 3500 meter buffer -2.00500 0.60230 2.294000 5.30400 0.86070
## 3250 meter buffer -2.44600 -0.85350 -0.697700 -4.48900 -0.64980
## 3000 meter buffer -2.13700 -0.80130 -1.454000 -8.13900 -1.22600
## 2750 meter buffer -1.95800 -1.51600 -1.743000 -10.23000 -1.66000
## 2250 meter buffer -1.75100 -0.54820 0.076010 0.62950 -0.64770
## 2000 meter buffer -1.28200 0.01076 -0.002506 0.43760 -0.64780
## 2500 meter buffer -1.49300 -1.10300 -1.262000 0.99500 -1.82600
## cnd(ppl) cnd(rds) cnd(ssm_lns) cnd(trl) cnd(wll) df logLik
## 250 meter buffer -0.58350 0.04315 -0.24170 -0.1009 -1.53600 15 -294.398
## 500 meter buffer -1.18100 -0.37350 -0.52020 -0.5640 0.49210 15 -313.151
## 1000 meter buffer 0.46020 0.66440 0.71130 0.6546 -1.01000 15 -313.579
## 750 meter buffer -0.33060 0.39990 0.08933 0.1862 -0.04885 15 -313.611
## 1500 meter buffer 1.03500 0.96360 1.27100 1.6530 -0.36790 15 -314.163
## 4000 meter buffer -0.05392 0.81850 0.80620 1.1670 -0.69150 15 -315.277
## 4250 meter buffer 0.33400 1.50500 1.15300 1.4560 0.31490 15 -315.758
## 1250 meter buffer 0.95440 0.90300 0.84580 1.2300 -0.93390 15 -315.859
## 4500 meter buffer 1.63800 2.26200 1.74200 1.8280 -0.70290 15 -316.302
## 5000 meter buffer -1.26300 1.42800 0.56130 0.5805 0.78170 15 -317.172
## 4750 meter buffer -0.63540 1.47000 0.79400 0.7656 0.83780 15 -317.404
## 3750 meter buffer 0.57060 1.07800 1.07100 1.3880 -0.48320 15 -317.432
## 1750 meter buffer 0.40550 0.03219 0.92360 0.6467 1.10800 15 -317.524
## 3500 meter buffer 0.72160 1.32200 1.38500 1.4750 0.14660 15 -318.383
## 3250 meter buffer 0.18050 0.74830 1.18900 1.1820 1.57800 15 -318.817
## 3000 meter buffer 0.46020 0.92300 1.24900 1.3140 2.00700 15 -319.422
## 2750 meter buffer 0.87090 0.47250 0.88000 0.8917 1.42000 15 -320.792
## 2250 meter buffer 0.99960 0.23030 0.74480 0.8051 0.13300 15 -321.218
## 2000 meter buffer 0.76620 0.27650 0.88490 0.7060 0.53110 15 -321.295
## 2500 meter buffer 0.59280 0.25370 0.97210 1.0170 1.66700 15 -321.567
## AICc delta weight
## 250 meter buffer 622.5 0.00 1
## 500 meter buffer 659.8 37.29 0
## 1000 meter buffer 660.7 38.14 0
## 750 meter buffer 660.8 38.21 0
## 1500 meter buffer 661.9 39.31 0
## 4000 meter buffer 664.1 41.54 0
## 4250 meter buffer 665.0 42.50 0
## 1250 meter buffer 665.2 42.70 0
## 4500 meter buffer 666.1 43.59 0
## 5000 meter buffer 667.9 45.33 0
## 4750 meter buffer 668.3 45.79 0
## 3750 meter buffer 668.4 45.85 0
## 1750 meter buffer 668.6 46.03 0
## 3500 meter buffer 670.3 47.75 0
## 3250 meter buffer 671.2 48.62 0
## 3000 meter buffer 672.4 49.83 0
## 2750 meter buffer 675.1 52.57 0
## 2250 meter buffer 676.0 53.42 0
## 2000 meter buffer 676.1 53.58 0
## 2500 meter buffer 676.7 54.12 0
## Models ranked by AICc(x)
## Random terms (all models):
## cond(1 | array)hmmmm seems fishy to me that the 250 meter buffer which is the only one that had missing data would perform THAT much better than all the others, and really you shouldn’t compare models if they aren’t run on the same data, hence the warning message
Let’s remove the 250 buffer and see what happens
black_bear_mods_no250 <- black_bear_mods %>%
purrr::discard_at('250 meter buffer')
# run model selection again
model.sel(black_bear_mods_no250)## Model selection table
## cnd((Int)) dsp((Int)) cnd(brr) cnd(lc_c11) cnd(lc_c20)
## 500 meter buffer -1.8100 + -8.8800 1.48300 2.1650000
## 1000 meter buffer -1.2440 + 5.0040 -0.79470 0.9692000
## 750 meter buffer -0.9882 + 3.2100 0.08085 1.0990000
## 1500 meter buffer -1.2430 + -1.6840 -1.97700 -0.1470000
## 4000 meter buffer -2.5810 + -10.0700 1.06300 2.4720000
## 4250 meter buffer -3.1640 + -12.3600 0.46190 3.0210000
## 1250 meter buffer -1.5400 + -3.2830 -0.87330 0.6002000
## 4500 meter buffer -3.1770 + -11.5900 -0.12570 2.3210000
## 5000 meter buffer -2.9500 + -9.5120 2.01100 3.0470000
## 4750 meter buffer -2.9020 + -10.2100 1.21200 2.9250000
## 3750 meter buffer -2.0630 + -7.5570 -0.28530 1.8680000
## 1750 meter buffer -1.2670 + -2.4830 -1.42300 0.1190000
## 3500 meter buffer -2.1010 + -7.2870 -0.70810 1.3820000
## 3250 meter buffer -0.3758 + -3.9270 -2.37000 0.0878400
## 3000 meter buffer -0.2031 + -3.0550 -2.93400 -0.1205000
## 2750 meter buffer 0.2775 + 0.8066 -2.37500 -0.2794000
## 2250 meter buffer -0.6461 + 2.6130 -1.06400 0.5436000
## 2000 meter buffer -0.7442 + 0.5012 -1.60200 -0.0001237
## 2500 meter buffer -0.1102 + 0.9909 -2.07700 -0.2207000
## cnd(lc_c21) cnd(lc_c22) cnd(lc_c23) cnd(lc_c34) cnd(lc_c50)
## 500 meter buffer 1.0230 2.56100 1.662000 2.7000 1.47400
## 1000 meter buffer -0.9961 0.63190 0.488000 -0.1981 0.25530
## 750 meter buffer -0.6852 1.29900 0.121600 -1.5010 0.01754
## 1500 meter buffer -1.1340 1.20900 -0.657200 -2.5990 -0.55680
## 4000 meter buffer -1.2820 1.08200 3.415000 31.6200 1.35600
## 4250 meter buffer -1.3010 2.05700 3.583000 26.0100 1.64500
## 1250 meter buffer -0.5675 1.28500 0.170100 0.4526 0.06635
## 4500 meter buffer -2.2420 1.93700 3.601000 -1.3670 1.68800
## 5000 meter buffer -1.2650 0.35690 5.558000 24.9100 2.14500
## 4750 meter buffer -1.7380 0.94160 4.978000 22.7400 1.84500
## 3750 meter buffer -1.8870 0.39800 2.642000 11.3000 1.01100
## 1750 meter buffer -1.0270 1.33500 0.063110 4.3790 0.10100
## 3500 meter buffer -2.0050 0.60230 2.294000 5.3040 0.86070
## 3250 meter buffer -2.4460 -0.85350 -0.697700 -4.4890 -0.64980
## 3000 meter buffer -2.1370 -0.80130 -1.454000 -8.1390 -1.22600
## 2750 meter buffer -1.9580 -1.51600 -1.743000 -10.2300 -1.66000
## 2250 meter buffer -1.7510 -0.54820 0.076010 0.6295 -0.64770
## 2000 meter buffer -1.2820 0.01076 -0.002506 0.4376 -0.64780
## 2500 meter buffer -1.4930 -1.10300 -1.262000 0.9950 -1.82600
## cnd(ppl) cnd(rds) cnd(ssm_lns) cnd(trl) cnd(wll) df logLik
## 500 meter buffer -1.18100 -0.37350 -0.52020 -0.5640 0.49210 15 -313.151
## 1000 meter buffer 0.46020 0.66440 0.71130 0.6546 -1.01000 15 -313.579
## 750 meter buffer -0.33060 0.39990 0.08933 0.1862 -0.04885 15 -313.611
## 1500 meter buffer 1.03500 0.96360 1.27100 1.6530 -0.36790 15 -314.163
## 4000 meter buffer -0.05392 0.81850 0.80620 1.1670 -0.69150 15 -315.277
## 4250 meter buffer 0.33400 1.50500 1.15300 1.4560 0.31490 15 -315.758
## 1250 meter buffer 0.95440 0.90300 0.84580 1.2300 -0.93390 15 -315.859
## 4500 meter buffer 1.63800 2.26200 1.74200 1.8280 -0.70290 15 -316.302
## 5000 meter buffer -1.26300 1.42800 0.56130 0.5805 0.78170 15 -317.172
## 4750 meter buffer -0.63540 1.47000 0.79400 0.7656 0.83780 15 -317.404
## 3750 meter buffer 0.57060 1.07800 1.07100 1.3880 -0.48320 15 -317.432
## 1750 meter buffer 0.40550 0.03219 0.92360 0.6467 1.10800 15 -317.524
## 3500 meter buffer 0.72160 1.32200 1.38500 1.4750 0.14660 15 -318.383
## 3250 meter buffer 0.18050 0.74830 1.18900 1.1820 1.57800 15 -318.817
## 3000 meter buffer 0.46020 0.92300 1.24900 1.3140 2.00700 15 -319.422
## 2750 meter buffer 0.87090 0.47250 0.88000 0.8917 1.42000 15 -320.792
## 2250 meter buffer 0.99960 0.23030 0.74480 0.8051 0.13300 15 -321.218
## 2000 meter buffer 0.76620 0.27650 0.88490 0.7060 0.53110 15 -321.295
## 2500 meter buffer 0.59280 0.25370 0.97210 1.0170 1.66700 15 -321.567
## AICc delta weight
## 500 meter buffer 659.8 0.00 0.331
## 1000 meter buffer 660.7 0.86 0.216
## 750 meter buffer 660.8 0.92 0.209
## 1500 meter buffer 661.9 2.02 0.120
## 4000 meter buffer 664.1 4.25 0.040
## 4250 meter buffer 665.0 5.21 0.024
## 1250 meter buffer 665.2 5.42 0.022
## 4500 meter buffer 666.1 6.30 0.014
## 5000 meter buffer 667.9 8.04 0.006
## 4750 meter buffer 668.3 8.51 0.005
## 3750 meter buffer 668.4 8.56 0.005
## 1750 meter buffer 668.6 8.75 0.004
## 3500 meter buffer 670.3 10.46 0.002
## 3250 meter buffer 671.2 11.33 0.001
## 3000 meter buffer 672.4 12.54 0.001
## 2750 meter buffer 675.1 15.28 0.000
## 2250 meter buffer 676.0 16.13 0.000
## 2000 meter buffer 676.1 16.29 0.000
## 2500 meter buffer 676.7 16.83 0.000
## Models ranked by AICc(x)
## Random terms (all models):
## cond(1 | array)this looks much more realistic
Now repeat for other species